Jan Kiszka wrote:
Hi,
when loading latency_rt.ko, e.g., and issueing a "cat /proc/rtai/sched",
you get a nice kernel oops (at least with latest CVS).
The reason: xnthread_user_pid(), as used by sched_read_proc and
xnpod_fatal_helper(), doesn't work with kernel-only threads. The
attached patch avoids this problem.
Jan
------------------------------------------------------------------------
Index: include/nucleus/thread.h
===================================================================
RCS file: /cvs/rtai/fusion/include/nucleus/thread.h,v
retrieving revision 1.29
diff -u -p -r1.29 thread.h
--- include/nucleus/thread.h 10 Jul 2005 11:30:40 -0000 1.29
+++ include/nucleus/thread.h 14 Jul 2005 10:50:36 -0000
@@ -202,7 +202,8 @@ typedef struct xnhook {
#define xnthread_signaled_p(thread) ((thread)->signals != 0)
#define xnthread_user_task(thread)
xnarch_user_task(xnthread_archtcb(thread))
#define xnthread_user_pid(thread) \
- (testbits((thread)->status,XNROOT) ? 0 :
xnarch_user_pid(xnthread_archtcb(thread)))
+ (testbits((thread)->status,XNROOT) || !xnthread_user_task(thread) ? \
+ 0 : xnarch_user_pid(xnthread_archtcb(thread)))
#ifdef __cplusplus
extern "C" {
Applied, thanks.
--
Philippe.
